🎗Impact of different eplet matching strategies on the feasibility and equity of eplet based kidney allocation

Introduction: Eplet based kidney allocation has the potential to improve long-term outcomes but the optimal eplet matching strategy remains uncertain. The objective of this study was to conduct a real-world analysis to assess the feasibility and equity of eplet based allocation by HLA DR, DQ, and DRDQ eplet mismatches.
Methods: We performed eplet matching on all active wait-listed kidney candidates (n=433) with each consecutive deceased donor (n=138) recovered between Nov 2022 – Nov 2023 in the province of British Columbia. To simulate real-world practice, eplet matching was only performed if the candidate was ABO-identical and virtual crossmatch negative (i.e. absence of donor-specific antibodies) with a donor. Eplet matching was determined using single molecule eplet mismatch at DR and DQ, where well-matched was defined as DR <7 and/or DQ <9. The ability of a candidate to find a well-matched donor was indicated by the proportion of ABO-identical/HLA compatible donors that had low eplet mismatch with the candidate.
Results: A total of 6249 eplet matches were performed between 433 candidates against 138 donors. Eplet matching was highly feasible with most donors finding at least one well-matched candidate by DR (96%) or DQ (85%), but this was more restrictive by DRDQ (68%) where 1 in 3 donors could not find any well-matched candidate on the waiting list (Figure 1). Irrespective of the eplet matching strategy, the top ranked candidate by eplets  in each allocation sequence was also ranked among the top by standard waiting times (median [IQR] rank: DR: 2 [1, 4], DQ: 3 [1, 7], DRDQ: 5 [2, 9]), suggesting little impact on wait time disparity with eplet based allocation. From the candidate’s perspective, the ability to find a well-matched donor was unrelated to their level of sensitization as measured by calculated panel of reactive antibody (cPRA) (p>0.05) or by candidate’s ethnicity (p>0.05, Figure 2). However, the candidate’s ability to match was highly dependent on the matching strategy – allocation by DRDQ resulted in half of the candidates being unable to find a well matched donor over one year (probability of finding a well-matched donor median [IQR]: 0% [0%, 15%])), but the ability to match improved significantly when considering DQ (17% [0%, 33%]) or DR (24% [0%, 50%]) alone (Figure 2).
Conclusions: The feasibility of eplet matching and the ability of candidates to find well-matched donors could be further enhanced when considering DR or DQ eplet alone compared with DRDQ eplet matching. Irrespective of the matching method, candidate ethnicity was not associated with their ability to find a well-matched donor. This real-world data could help inform the design of an eplet based allocation system that balances efficiency and equity.